Volume-Outcome Relationship

Background

Training and repetition are necessary to learn the skills needed to expertly accomplish a surgical procedure or become familiar with protocols and organizational nuances in any particular hospital setting. However, the “practice makes perfect” hypothesis raises a series of questions when applied to the real world. How high is the threshold necessary to acquire competency? Once one achieves that threshold, for example, through rigorous training, do the skills deteriorate over time if not maintained? Does quality continue to get better with experience (or volume) above the threshold—that is, should one seek out the highest-volume provider or just avoid those below a threshold level? For surgical procedures, is it just the volume of the primary surgeon, or do the skills of the anesthesiologist and other members of the team matter? In a set of procedures, such as coronary artery bypass graft (CABG), does volume matter for all cases or just for a subset of cases, such as the most risky patients or cases when an unexpected event occurs? An entirely different perspective on the simple association between volume and outcome is that the conventional wisdom is backward. That is, perhaps some physicians are just better than others and receive more referrals because of their better outcomes and thus have higher volumes; this is known as the selective-referral hypothesis. If so, are there subtle techniques and protocols that can be taught so that others with lower case volumes can also achieve better outcomes? There is no reason, moreover, to believe that both practice makes perfect and selective referral may not occur simultaneously, perhaps with differential importance for various conditions and procedures.

Harold Luft and colleagues' 1979 article in the New England Journal of Medicine was the first to examine the volume-outcome relationship across a series of surgical procedures. This study examined the 1974–1975 discharge data from 1,498 hospitals on 12 surgical procedures. A volume-outcome relationship was observed for certain procedures, including open-heart surgery, coronary artery bypass, and vascular surgery, in which high-volume (defined as more than 200 procedures per year) [Page 1174]hospitals were associated with significantly lower mortality. However, for other services, such as colectomy and hip replacement, mortality also decreased with increasing hospital volume but stabilized at a much lower volume, between 10 and 50 procedures per year. Other procedures, such as cholecystectomy and vagotomy, showed no relation between volume and outcome.

Over the next 30 years, hundreds of studies in the health services research and clinical literature surfaced confirming the volume-outcome relationship for both hospitals and individual providers, although the evidence is stronger for the former. Those procedures and conditions that have been most studied include vascular surgery, cancer, and cardiac care. Important questions have surfaced regarding the volume-outcome relationship: What constitutes adequate volume, and how is this determined? Which procedures are the most sensitive to volume? To what extent is hospital volume—as opposed to physician volume—the key variable? How does one account for severity of illness of patients? Might there be selective biases in referral patterns? To what extent does accumulated experience, versus volume (or “throughput”) at a given point in time, account for good outcomes? What are the clinical and other implications of various policies potentially derived from the observed relationship? After three decades of work, these questions remain at the heart of volume-outcome research.

...